Main Project Image
Main Project Image
Main Project Image

All your AI assistants in one place.

Category

UI/UX

Team

Kevin Z (Developer)

My Role

Lead UI/UX Designer, Design System

Tools

Figma

Introduction

Nexus is an AI assistant service that enables users to seamlessly switch between multiple AI models mid-conversation with transparent on-demand pricing.

Originally a developer-initiated project, I was brought on to design the entire user experience, interaction flow, and branding.

Nexus is an AI assistant service that enables users to seamlessly switch between multiple AI models mid-conversation with transparent on-demand pricing.

Originally a developer-initiated project, I was brought on to design the entire user experience, interaction flow, and branding.

Nexus is an AI assistant service that enables users to seamlessly switch between multiple AI models mid-conversation with transparent on-demand pricing.

Originally a developer-initiated project, I was brought on to design the entire user experience, interaction flow, and branding.

Research

Understanding User Painpoints

To understand our users’ needs, we conducted interviews across three key groups: frequent, average, and seldom users. Our research revealed three major challenges:

Limited Value

Subscription pricing overcharges infrequent users for a few messages while making multi-model access costly for power users.

Too Many Accounts

Switching between multiple accounts and subscriptions across different AI platforms just to access various AI models is cumbersome.

Restricted Exploration

The subscription cost discourages users from experimenting with different models, limiting flexibility and engagement with AI services.

Market Research

To understand the competitive landscape, we analyzed existing AI services and identified three common business models.

Multiple AI Models, No Subscription

Allows users to chat with multiple AI models but require them to bring their own API keys for access.

Multiple AI Models, Requires Subscription

Offers access to various models but locks premium features behind a subscription.

Proprietary AI Model with Subscription

Limits users to AI models from a single provider and premium features locked behind a subscription.

Design Goals

Backed by research, we've identified key goals that our product should aim to embody

Seamless Model Switching

Switching AI models should be effortless; we should make sure users can compare responses and refine conversations without friction.

Transparent Pricing

Users should always have a clear breakdown of costs, with no hidden fees or confusion—just full control over what they pay for.

Intuitive & Logical Interface

A clean, responsive, and accessible UI that prioritizes usability and efficiency.

Design

Testing & Iterations

Our design process involved multiple iterations and usability tests to refine the interface based on user feedback. Early wireframes focused on layout structure and feature accessibility, while later versions improved navigation, model switching, and clarity.

Before and After

Our initial design prioritized simplicity, but user testing revealed key usability challenges. Below are key before-and-after improvements that improve transparency, functionality, and usability.

Before

  1. Overly Simplified Pricing – Users lacked clarity on exact cost calculations.

  2. Cheeky Token System – The gamified approach felt out of place for a transparent tool.

  3. Balance Anxiety – Displaying total balance near cost made users hesitant to spend.

After

  1. Detailed Cost Breakdown – Input/output pricing and contextual message count are clearly shown.

  2. Removed Token System– Uses direct dollar amounts for transparency.

  3. Separated Balance Display – Moved balance display next to the user profile to reduce cost-related anxiety and shift focus to per-message pricing.

Before

  1. Limited AI Model Control – Users couldn’t easily switch models after receiving a response.

  2. Overly Minimalist – Lacked basic features like copy, regenerate, and view message details.

  3. Unnecessary information Displaying cost and time per message directly didn’t add everyday value.

After

  1. Added Model Switching & Comparison – Users can now quickly compare and switch between models with a click or two.

  2. Quick Action Features – New buttons and toggles provide more features for the users.

  3. Condensed Unnecessary Information – Removed redundant details for a cleaner, more focused UI.

We translated our research findings and user feedback into these high-fidelity prototypes.

Large Project Gallery Image #1
Large Project Gallery Image #1
Large Project Gallery Image #1
  • The image featured in the carousel #1
  • The image featured in the carousel #2
  • The image featured in the carousel #3
  • The image featured in the carousel #4
  • The image featured in the carousel #1
  • The image featured in the carousel #2
  • The image featured in the carousel #3
  • The image featured in the carousel #4

Outcome

The Nexus MVP web app and official website is nearly complete and currently in the development handoff and QA stage. The developer and I are working closely, optimizing design for code and resolving any development challenges.

Next Steps

  • QA Testing – Refining interactions and fixing issues.

  • Beta Testing – Collecting user feedback for further improvements.

  • Feature Updates – Expanding functionality based on insights.

Vision

  • A powerful, all-in-one AI platform simplifying multi-model access.

  • An accessible and affordable AI service to everyone

  • Encouraging users to explore the capabilities of various AI models.

Takeaways

Throughout the design and development of Nexus AI, I learned a lot through trial and error and constructive feedback.

Developer Collaboration

Open communication with developers helped me understand best practices for optimizing design for code and improving the designer-developer workflow.

Prioritization Matters

Some features, while valuable, can be postponed due to resource constraints, allowing for a more efficient MVP.